INDEX

Explanations

words related to conflicts, violence, and international affairs

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

Data Type

torch.float32

Hook Point

blocks.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Trog

-0.66

arde

-0.64

 footing

-0.61

 engagements

-0.59

ennes

-0.59

 disapprove

-0.59

 communications

-0.57

 extrad

-0.57

 engagement

-0.56

 equip

-0.56

POSITIVE LOGITS

ÂłÂłÂłÂł

1.23

ÂłÂłÂłÂłÂłÂłÂłÂł

1.21

ÂłÂłÂł

1.17

ÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂł

1.16

Article

1.00

Advertisement

1.00

advertisement

0.95

ÂłÂł

0.91

Anyway

0.89

Nevertheless

0.87

Activations Density 1.117%

words related to conflicts, violence, and international affairs

No Comments

No Known Activations